An Association Based Query Expansion Approach Using Linked Data

Dahir, Sarah; Qadi, Abderrahim El; Bennis, Hamid

doi:10.1109/isivc.2018.8709216

Cited by 2 publications

(13 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This section details our proposal, which is an improved variant of our previous method "cosinesimilarity (COS-SIM) on linked vectors" [9]. In particular, we proposed two approaches called Label-based modified Concept2vec Approach and abstract-based modified Concept2vec approach.…”

Section: Methodsmentioning

confidence: 99%

“…If there is not any, we use the first subject; − Use of the SPARQL Protocol and RDF Query Language to find a list of concepts that have a determined dct: subject in common. It is the SPARQL [9] approach; − Comparison between the lists of concepts (previous step) that are ralted to query conepts with those related to the top documents using cosine similarity to determine expansion concepts. It is the "COS-SIM on linked vectors" [9] approach; − Use of the SPARQL [9] lists of concepts as well as the "COS-SIM on linked vectors" [9] expansion concepts as data for the modified Concept2vec model; − Creation of the skip-gram model using "(3) and ( 4)".…”

Section: Methodsmentioning

confidence: 99%

“…It is the SPARQL [9] approach; − Comparison between the lists of concepts (previous step) that are ralted to query conepts with those related to the top documents using cosine similarity to determine expansion concepts. It is the "COS-SIM on linked vectors" [9] approach; − Use of the SPARQL [9] lists of concepts as well as the "COS-SIM on linked vectors" [9] expansion concepts as data for the modified Concept2vec model; − Creation of the skip-gram model using "(3) and ( 4)". The number of context words we are looking at is 5 concepts before the input and 5 concepts after it; − Evaluation of the similarity between the initial query concepts and the "COS-SIM on linked vectors" expansion concepts from DBpedia using the following line on python: Similarity('[QueryConcept]','[COS-SIMonlinkeddataConcept]').…”

Section: Methodsmentioning

confidence: 99%

“…Dahir et al [9] proposed an approach, which is designed to annotate feedback documents of the initial query using DBpedia. Then, for each annotated DBpedia entity, the relevant "dct: subject" (an entity of type Thing) is adopted to find all entities having this subject as one of the "dct: subject" attribute's values using SPARQL protocol and RDF query language (SPARQL).…”

Section: None Use Of External Sourcesmentioning

confidence: 99%

“…To resolve this issue, we propose a new query expansion method that relies on DBpedia concepts to generate the vectors. Our approach extends our previous work on association-based query expansion [9]. And instead of using cosine similarity only at the end of the embedding process as an evaluation metric of the embeddings' quality as in [10]; we use it before the embedding to insure having only semantically related concepts from the start in the training data.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Query expansion based on modified Concept2vec model using resource description framework knowledge graphs

Dahir

Qadi

2023

IJ-AI

View full text Add to dashboard Cite

The enormous size of the web and the vagueness of the terms used to formulate queries still pose a huge problem in achieving user satisfaction. To solve this problem, queries need to be disambiguated based on their context. One well-known technique for enhancing the effectiveness of information retrieval (IR) is query expansion (QE). It reformulates the initial query by adding similar terms that help in retrieving more relevant results. In this paper, we propose a new QE semantic approach based on the modified Concept2vec model using linked data. The novelty of our work is the use of query-dependent linked data from DBpedia as training data for the Concept2vec skip-gram model. We considered only the top feedback documents, and we did not use them directly to generate embeddings; we used their interlinked data instead. Also, we used the linked data attributes that have a long value, e.g., “dbo: abstract”, as training data for neural network models, and, we extracted from them the valuable concepts for QE. Our experiments on the Associated Press collection dataset showed that retrieval effectiveness can be much improved when a skip-gram model is used along with a DBpedia feature. Also, we demonstrated significant improvements compared to other approaches.

show abstract

Section: Methodsmentioning

confidence: 99%